[BiO BB] Please guide me

Dan Bolser dmb at mrc-dunn.cam.ac.uk
Fri Jan 30 15:45:33 EST 2004

++ Michael Gruenberger--
> I can see why the idea of a common language is a good idea, but I don't understand
> how a schema would achieve that?
> If you want databases to talk to each other in a common language then you need a
> common network transport (e.g. web services, XML, SOAP, BioMoby) and a common set
> of terms or ontologies (e.g. GO, EMAP, MA, MPATH, see http://obo.sourceforge.net/
> or http://www.geneontology.org).

Thanks for the link.

> This would basically allow your databases to talk to each other and I'm sure a
> common schema would make life easier, but it would be a lot of additional work to
> integrate this 'global schema' into existing
> databases. And some if not most of it would be redundant, at least for our
> database, because we don't for example store information about genes, but simply
> store the gene name + GO id and then link to other database which have more
> detailed information.

Schemas (or more generally data models) provide context for data. Rather than simply
enumerating the symbolic links between data, a data model hopes to organize the data
conceptually, providing semantic meaning to abstract objects. You are absolutely
right that a 'global schema' is more than most people need and is hard to agree on
or implement. What I mean is a bunch of independent schema's (schemata?) which can
interoperate through common terms (ontologies as you put it). No one researcher can
ever decide on the data model for his data, because making a model implies
understanding the field - which negates the process of research. This is why each
researcher should develope independent sub models (modlets?) for the data they work
on, within a machine readable format to allow computer scientists to interoperate
between schemata.

> Hmmm... I'd really like to understand your idea though! Could you give an example
> of how this would work?

In the simplest terms, I imagine a wet lab working on mitochondrial transporter
proteins. Because of their work, they develope a database of mitochondirial
transporter proteins. Then are (probably) the most qualified people in the world to
make this database, so (much in the way of DAS), they annotate their database as -
proteins - mitochondrial - transporters ... within the framework I imagine, that is
all that is needed - ontological mappings allow this data to be integrated into a
'global schema', but it is up to the individual researcher as to which labs (data
sources) he trusts, and eventually a common consensus will emerge. This is the way
science works.

Sorry for the slightly garbled response,

> Cheers,
> Michael.
> --
> Michael Gruenberger
> Database developer, Pathbase, http://www.pathbase.net
> PGP-Public Key ID: 278E1DFF
> On Fri, 2004-01-30 at 09:34, Dan Bolser wrote:
>> ++ HKG--
>> >>> What would the benefits of a common schema be?
>> >
>> >>interoperability, foundation of the field, common development
>> >>environment, growth
>> >
>> > A common schem wont help one to achive that.
>> Not directly no - but it would be a big help!
>> Maybe this is a big prejudice of mine, but I am actually quite passionate about
>> this idea. After all, when one mathematician talks to another, they have a
>> common formal language with which to do so - chemists too and the same with
>> comp-sci. Now, while every biologist damn well knows he has a common language to
>> communicate with his peers, this language is not easily formalized. The previous
>> examples are able to work by defining primitives and association rules, but in
>> biology the primitives are already high level concepts, and the association
>> rules are often the results of ongoing research activities. Thus these
>> 'fundamentals' are not easy to agree - this is why we must develop a formal
>> framework within which precise definitions and relationships can emerge
>> pragmatically, in effect leading to a global schema. From the viewpoint of the
>> philosophy of science I think this is how it works in the brains of biologists
>> (I could be wrong), the difficulty is getting something similar to happen in a
>> computer friendly way. I am 100% confident (maybe you see my prejudice) that
>> this will happen someday.
>> Ta,
>> Dan.
>> >>I guess people can add more.
>> > _______________________________________________
>> > BiO_Bulletin_Board maillist  -  BiO_Bulletin_Board at bioinformatics.org
>> https://bioinformatics.org/mailman/listinfo/bio_bulletin_board
>> _______________________________________________
>> BiO_Bulletin_Board maillist  -  BiO_Bulletin_Board at bioinformatics.org
>> https://bioinformatics.org/mailman/listinfo/bio_bulletin_board

More information about the BBB mailing list